Every enterprise Salesforce org has "the attic"—that dark, dusty corner of ancient custom objects, thousands of unused fields, and undocumented code that everyone is absolutely terrified to touch. But here is the hard truth: if your metadata is a chaotic mess, your autonomous AI agents will be completely disoriented.

In Session 2 of De-Fragging the Franken-Org, we focus our gaze entirely through the Metadata Lens to determine if your configuration is telling the truth or hiding years of broken architectural choices. AI agents rely heavily on clear metadata, field descriptions, and clean structures to understand context and execute user intent. If your metadata is misleading, your Agentforce implementation will fail. In this technical session, we will open up the attic and show you exactly how to untangle the web.

The Core Question: Is our configuration telling the truth?

What we’ll cover:

• The AI Intent Gap: How missing descriptions and dead code confuse LLMs and agents.

• Metadata Decluttering: Strategies to safely deprecate components without breaking dependencies.

• Reading the Blueprint: Auditing custom settings, labels, and configurations for architectural alignment.

Live Tool Deep Dive:

• Metadata API: Peeling back the layers of your configuration programmatically.

• SFDX Scanner: Running static code and architecture analysis to sniff out bad code smells.

• Org Browser: Navigating your org’s metadata directly from VS Code to find hidden skeletons.

Stop guessing what your configuration does. Join us to learn how to clean out the metadata attic so your future AI agents have a clear, uncompromised map of your business logic.